Clustering citation histories in the Physical Review

作者：

Highlights：

• A clustering procedure involving Spectral clustering is proposed in order to find clusters of similar articles according to their citation history.

• The method is applied to the full Physical Review dataset.

• Two main article archetypes are found and discussed: marathoners and sprinters.

• Multinomial logistic regression is used to find determinants of cluster membership.

摘要

Highlights•A clustering procedure involving Spectral clustering is proposed in order to find clusters of similar articles according to their citation history.•The method is applied to the full Physical Review dataset.•Two main article archetypes are found and discussed: marathoners and sprinters.•Multinomial logistic regression is used to find determinants of cluster membership.

论文关键词：Citation histories,Clustering,Regression analysis,Physical Review

论文评审过程：Received 31 May 2016, Revised 4 July 2016, Accepted 9 July 2016, Available online 20 September 2016, Version of Record 20 September 2016.

论文官网地址：https://doi.org/10.1016/j.joi.2016.07.009